Page 1
© Professor Rifat Atun. Imperial College London, 2008
Positive Synergies between Global Health Initiatives
Professor Rifat AtunProfessor of International Health Management, Imperial College London &Director Strategy, Policy and Performance Cluster, The Global Fund to Fight AIDS, TB & Malaria
Page 2
© Professor Rifat Atun. Imperial College London, 2008
Key Research Questions
• What are the extent and nature of integration of GHIs (programmes they finance) and health systems to achieve synergies in varied contexts?
• Which factors influence the extent and nature of integration?
• How the varied health system designs and delivery structures influence outcomes?
Page 3
© Professor Rifat Atun. Imperial College London, 2008Atun, Ohiri, Adeyi, 2008
Key variables affecting the nature and extent of integration
1. The Problem being addressed
2. The Intervention
3. The Adoption System
4. The Health System characteristics
5. The Broad Context
Page 4
© Professor Rifat Atun. Imperial College London, 2008
Integrate or not to integrate: framework for analysis
InterventionAdoption
System
Broad Context
Broad Context
Health System Characteristics
Problem
Page 5
© Professor Rifat Atun. Imperial College London, 2008
The Problem
• Necessity and Urgency • Burden– Economic and social
• Perceived and real• Social Narrative
• Transmission dynamics
Page 6
© Professor Rifat Atun. Imperial College London, 2008
The Intervention
• Complexity
• Scalability
• Simpler to more complex*
• Replicability
* See next slide
Page 7
© Professor Rifat Atun. Imperial College London, 2008
Intervention: simple versus complex
Single episode
Multiple episodes
Few elements
Multipleelements
Less complex
More complex
Atun and Kyratsis 2007
Page 8
© Professor Rifat Atun. Imperial College London, 2008
Intervention: simple versus complex
Few stakeholders
Multiple stakeholders
Few levels
Multiplelevels
Less complex
More complex
Atun and Kyratsis 2007
Page 9
© Professor Rifat Atun. Imperial College London, 2008
Intervention: simple versus complex
User engagement lower
User engagement higher
Technologydominates
Less complex
More complex
Atun and Kyratsis 2007
Behaviourdominates
Page 10
© Professor Rifat Atun. Imperial College London, 2008
The Adoption System
• Receptivity • Individual & organisational
• Political economy• Incentives
– agency/provider/user incentive alignment
• Legitimacy– Cognitive
– Technical
– Normative
– Economic
Page 11
© Professor Rifat Atun. Imperial College London, 2008
Health System Characteristics
• Feasibility • Governance
• Structure and organization
• Financing
• Provider payment methods
• Resource availability
• Service delivery
• M&E system
Page 12
© Professor Rifat Atun. Imperial College London, 2008
The Context
• Sustainability
• Attributability
• Fiscal space– Overall and health sector
specific
• Frailty
• Reporting needs
Page 13
© Professor Rifat Atun. Imperial College London, 2008
The Context
• Opportunity
• Desirability
• Critical events– Visibility
• Synergy
• Technology / innovation
• Political economy
• Socio-cultural factors
Page 14
© Professor Rifat Atun. Imperial College London, 2008
Analysing the extent and nature of integration
Page 15
© Professor Rifat Atun. Imperial College London, 2008
Integration into Critical Health System Functions
1. Governance– Reporting– Accountability
2. Financing– Pooling– Provider payment
3. Planning – Needs assessment– Priority setting – Resource allocation
Page 16
© Professor Rifat Atun. Imperial College London, 2008
Integration into Critical Health System Functions
4. Service Delivery– Structural
• Human resources, • Shared infrastructure
– Operational integration• Supply chain• Guidelines• Procurement
5. Monitoring and Evaluation– Data collection and analysis
6. Demand Generation– Financial incentives – e.g. CCT, insurance– Population interventions – e.g. education and promotion
Page 17
© Professor Rifat Atun. Imperial College London, 2008
Some Initial Results
Page 18
© Professor Rifat Atun. Imperial College London, 2008
Intervention Complexity
Few Many
Single
Multiple
Dengue - Cuba Malaria - Colombia
Schistosomiasis - Brazil, Burundi, Cameroon, China, Saudi Arabia, Uganda
Leprosy - India, Sri Lanka
Onchocerciasis - Uganda
Nutrition - Peru, etc.
IMCIICDS
Dular - India
FP/MCH - Matlab, Bangladesh
FP/MCH - Pakistan - LHWPFP/MCH - Nepal (Tuladhar)
HIV/AIDS - Haiti
STD - Mbofana
FP; STD - Lafort
Mental health - WhettenSubstance abuse - Friedmann
Intervention
elements
Intervention frequency/number of episodes
Page 19
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Fully integrated
Partially integrated
Not integrated
? Unknown
Most to all outcomes
Mixed outcomes
Few to no outcomes
? Unknown
Page 20
© Professor Rifat Atun. Imperial College London, 2008
Page 21
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Cuba (ToledoRomani2007)
Colombia (Rojas2001)Malaria
Dengue
Service deliv
ery
Finance
Governance
Demand generatio
n
Monitorin
g & E
valuation
?
Planning
Page 22
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Governance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
Schistosomiasis control
Burundi (Engels1993,1995)
Cameroon (Bausch1995,Cline1996)
China (Sleigh1998)
Saudi Arabia (Ageel 1997)
Brazil (Filho1992)
?
?
?
?
?
?
?
?
?
?
?
Uganda (Kabatereine 2006)
Page 23
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Steward
ship/G
overnance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
India (Rao 2002, Thakar 2003)
Leprosy
Sri-Lanka (Kasturiaratchi 2002)
??
Page 24
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Governance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
Peru
Nutrition
Bangladesh (Hossain2005) ??
Various (Deitchler2004) ?
Page 25
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Governance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
IMCI*
ICDS - India (Agarwal2000, Kapil1999)
Child health & development
Dular - India (Dubowitz2007)
?
?
?
Page 26
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Governance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
Bangladesh – FPHSP (Philips1984, de Graff 1986)
Pakistan – LHWP (Douthwaite 2005)
Family Planning services
Nepal (Tuladhar 1982)
?
?
?
?
?
?
?
?
?
Page 27
© Professor Rifat Atun. Imperial College London, 2008
Extent of integration & success as documented in studies
Success
Service d
elivery
Finance
Steward
ship/G
overnance
Demand g
eneratio
n
Monitorin
g & E
valuatio
n
Planning
Haiti (Peck 2003)
HIV/AIDS & STD services
? ??
Page 28
© Professor Rifat Atun. Imperial College London, 2008
Conclusions
1. Extent and nature of integration varies
2. Context matters: complex adaptive systems at play
3. Reductionist approaches counterproductive: aim to ‘unpack’ what is meant by integration
Page 29
© Professor Rifat Atun. Imperial College London, 2008
Empirical Research
Page 30
© Professor Rifat Atun. Imperial College London, 2008
Case Study Approach
• Exploratory
• Descriptive
• Explanatory
Page 31
© Professor Rifat Atun. Imperial College London, 2008
Design
Logic of design key
Single Multiple
Holistic
Embedded
Russia TB
Estonia PHC
Russia HIV
Africa HIV
Euro PHC
Tech adoption
Baltic PPP
Page 32
© Professor Rifat Atun. Imperial College London, 2008
Embedded units
• HIV
• TB
• Malaria
• NTDs
• Regions
Page 33
© Professor Rifat Atun. Imperial College London, 2008
Analytic vs. Statistical Generalisation
• Cases not sampling units but each akin to an individual ‘experiment’
• Analytic generalisation using theory developed a prioriReplication logic– n number of case studies support the same theory– n number of case studies do not support a rival theory
• Statistical generalisationSampling logic
Page 34
© Professor Rifat Atun. Imperial College London, 2008
Careful case selection
• Literal replication– Each predict similar results (n=4)
• Theoretical replication– Predict contrasting results --- but for predictable
reasons (n=4)
Page 35
© Professor Rifat Atun. Imperial College London, 2008
Theoretical framework and propositions key
• State the conditions under which particular phenomena are likely to be found– Allows literal replication
• State the conditions when particular phenomena are not likely to be found– Allows theoretical replication
Page 36
© Professor Rifat Atun. Imperial College London, 2008
Closed vs. flexible design
• Closed but with inductive analysis– Retain replication logic– Build theory as an output– Test ‘additional’ new/alternative propositions
• Flexible and inductive– Risk of drift
Page 37
© Professor Rifat Atun. Imperial College London, 2008
One or two tail design
• Good outcome
• Good and poor outcome
Page 38
© Professor Rifat Atun. Imperial College London, 2008
Data
• Mixed methods
• Multiple sources
• Inductive
• Iterative
• Triangulation
Page 39
© Professor Rifat Atun. Imperial College London, 2008
Process
Agree theory
Generate
propositions
Analytical tools
approach & d/base
Pilot cases
Case studies Case studies
Refine tools
Literal replication Theoretical replication
Explanatory theory & Evidence
Propositions Rival propositions
Construct validity Reliability
Construct validity
Internal validity
Internal validity External validity
Page 40
© Professor Rifat Atun. Imperial College London, 2008
Cases
Africa
• Tanzania
• Ghana
SE Asia
• Thailand
• Viet Nam
Embedded units of analysis
• NTDs + malaria + TB + HIV
• NTDs + malaria + TB + HIV
• Malaria, TB, HIV?
• Malaria, TB, HIV?